Paseo plugin that classifies each message with Jev or a local Laya model and routes it to one of your presets, each carrying its own provider, model, reasoning setting and instructions.
obetomuniz/auto-mode-for-paseo
User repository on GitHub
Paseo plugin that routes each message to a preset on Codex, Claude, OpenCode, or any other installed provider.
Repo stats from the GitHub API, cached Oct 2026.
Auto Mode for Paseo is a Paseo plugin. It reads each new message and sends it to the best preset for the task. Each preset has its own provider, model, reasoning setting, and instructions. A preset can use Codex, Claude, OpenCode, or another provider installed in Paseo. TypeSafe Jev or a local Laya model classifies the message. The OpenAI Decisions API is a third, experimental classifier.
Quoted from the project's README. Read the full README.
Watch out
The README opens by calling this an experiment and saying not to use it in production: the classifier only picks a preset, it does not read your code or plan the work, so it can choose wrong, which is why it tells you to examine each routing notice and select manually when the choice matters. Its own numbers are smoke tests, not a benchmark — with the default presets plus two custom ones Jev picked the expected preset in 16 of 18 requests and Laya in 14 of 18, scope tagging was 12 of 12 against 10 of 12, and the routing tests validate selection rules rather than accuracy. Laya's quality for this task is explicitly unbenchmarked and its confidence is not a guarantee of intent. Requirements are specific: Paseo 0.8.0 or later, Node.js 24, and an installed, authenticated Paseo provider for every enabled preset. Copy reviewed against the repository README on 2026-10-04, not from running it.
rhighs/jev-codeTypeScript
kushals256/jevcacheTypeScript
rajdhakad9826/jev-routerTypeScript
muratcakmak/jev-guardTypeScript
tonyzdev/pijevTypeScriptPosts from builders shipping with Jev right now.
